Basic economics: when something becomes abundant, its value falls. When something becomes scarce, its value rises. AI has made content production abundant — anyone can produce any text in seconds. Well-written, coherent, informative content is no longer a differentiating advantage. It is the norm. What remains scarce — and whose value is rising — is precisely what AI cannot produce: the irreproducible lived experience, the authentic contradiction, the error that reveals thought, the conviction carried by a real story.

This mechanism is exactly the one the corpus documented in "The Infrational Loop" and "The Infrational Crisis" — but applied to content rather than technical skills. DevOps diluted bare-metal competences by making them vague and universal → real bare-metal competence became scarce → its value rose. AI dilutes content production by making it universal → authentic human content becomes scarce → its value rises. Same mechanism. Two different domains. The corpus documented infrastructure. Sirine documented content. The thesis is the same.

AI can write about loss, resilience, failure, joy. It can do so with precision and even a certain elegance. What it cannot do is have lived through these experiences. The difference is not stylistic — it is ontological. When Sirine writes about her journey at the intersection of tech, marketing and human sciences, she mobilises an accumulation of real experiences, imperfect choices, encounters, errors and reconsiderations spanning years. This trajectory cannot be reproduced by a language model — because it was never data to train on. It is unique because it was lived, not because it is well written.

An LLM optimises for coherence. It produces internally consistent texts without contradictions, without position reversals, without admissions of past error. This is precisely its weakness — because living thought is recognised by its capacity to contradict itself, to evolve, to acknowledge it was wrong. Contradiction is not a defect of human thought — it is its signature. An author who has never contradicted themselves has not thought much. An author who visibly evolves over years creates an intellectual trajectory that nobody can imitate without having genuinely lived through the same experiences.

The dyslexic's spelling errors are more authentic than the model's perfect text. This is not a defence of carelessness — it is an observation about authenticity signals. Embodiment is the presence of a real person in what they produce — their hesitations, their recurring formulas, their emphases, their avoided subjects, their compulsive returns to the same themes. It is the corpus of a life that shows through in the corpus of a work. AI produces texts without a body, without history, without obsessions. It cannot embody itself because it has no body to embody.

For a long time, style was an authenticity signal. A well-written, structured, argued text signalled an author who mastered their subject and had taken the time to formulate their thought. This signal is now neutralised. An LLM produces well-written, structured and argued texts in seconds. Style is no longer a reliable filter. The question "who wrote this?" is legitimate for any text — including this one. This is not an AI crisis. It is an epistemological crisis of content.

Sirine identifies the shift that is taking place: we will no longer follow experts — we will follow people. The distinction is fundamental. The expert is certified on what they know — a declarable, reproducible, transferable competence. An AI can simulate expertise. The person is reliable on what they are — a life trajectory, a coherence between words and actions, a presence in time that cannot be simulated over years. Reliability is not a certification — it is an observable accumulation.

In a world where style no longer filters, what trust signals remain?

— Temporal coherence: someone who has been saying the same thing for five years — with documented evolution — cannot be a bot. Duration is an authenticity signal that AI cannot retrospectively simulate.
— Physical and social presence: people met, events attended, photos existing in verifiable contexts.
— Irreducible specificity: details too precise to be invented — a colleague's name, a restaurant's address, the time of a meeting that went wrong.
— Assumed error: someone who acknowledges having been wrong, who documents their change of position, who exposes themselves to criticism by showing their contradictions.

Sirine formulates the distinction precisely: "coherent people (who do what they say, not who say what they don't do)". This formula distinguishes two types of coherence — declarative coherence (I say I do X) and operational coherence (I have actually done X, here are the traces). Operational coherence is verifiable, documentable, difficult to simulate. It takes time to build and no time to destroy. It is the same principle as reputation in cryptography: difficult to acquire, easy to lose.

Amine RAITI has been writing about bare-metal infrastructure for years. He trained students from 2006. He worked at Ecritel and Weborama. The former students he meets again at major companies are living proof that what he says corresponds to what he did. This is not declared expertise — it is operational coherence accumulated over two decades. An AI can imitate his writing style. It cannot produce the engineers who came out of his classrooms.

One does not decide to be singular. One becomes singular through accumulation of real experiences, assumed choices, acknowledged errors, positions defended under pressure. Singularity is the irreducible residue of a lived life — not a content strategy. What makes Sirine singular is not that she decided to be at the intersection of three disciplines. It is that she arrived there through a series of real choices whose traces she carries. This residue, AI cannot generate. It can imitate it in one sentence. Not over ten years.

Cloud abstracts infrastructure → the engineer mastering bare-metal becomes scarce → their value rises.
AI abstracts content production → the human mastering their lived experience becomes scarce → their value rises.

In both cases, abstraction reveals what it cannot capture. Cloud cannot capture the tacit knowledge of the SRE engineer who hears an anomaly in fan noise. AI cannot capture Sirine's life trajectory at the intersection of three disciplines. These two irreducibles — tacit knowledge and embodied experience — are precisely what takes value when everything else is abstracted.

"The Infrational Crisis" documented how vocabulary dilutes technical professions — "DevOps engineer" instead of "systems administrator", "cloud engineer" instead of "network engineer". The same dissolution occurs in content: the "content creator" who uses AI to produce everything is no longer a creator — they are an assembler. The dissolution of the infrastructure body through title dilution and the dissolution of the creative body through AI delegation are two manifestations of the same phenomenon: when the function is outsourced, the competence that exercised it disappears.

AI as amplification: it takes Amine's ideas — forged over twenty years of field work, training, production incidents and reflection — and structures them, enriches them with documentary references, formats them into 7 imperial pages. AI brings formatting and rapid research. Amine brings the theses, the experiences, the positions, the nuances. The result is stronger than what Amine would have produced alone — and radically different from what AI would have produced without him.

AI as substitution: it produces the text instead of the human, who only validates or corrects. The result may be stylistically correct. It has no trajectory behind it. It has no lived experience. It has no Ecritel, no Weborama, no 2006 classroom, no former students met again in major companies. It is hollow — and the reader feels it, even if they do not know why.